Defective battery piece detection system and method

Abstract

The embodiment of the invention provides a defective battery piece detection system and method, and relates to the technical field of machine vision. The system comprises a conveyor belt, a laser, a camera and a processing unit, the laser is located on one side of a detection area in the conveyor belt, and the camera is located on the other side of the detection area; the conveying belt is used for conveying the to-be-detected battery pieces on the conveying belt; the laser is used for emitting laser to the detection area when the battery piece to be detected is conveyed to the detection area; the camera is used for collecting a to-be-detected image of the detection area under the irradiation of the laser emitted by the laser when the to-be-detected battery piece is conveyed to the detection area; sending the to-be-detected image to a processing unit; and the processing unit is used for detecting the to-be-detected battery piece to obtain a detection result about whether the to-be-detected battery piece has defects or not. The efficiency of defective battery piece detection can be improved, and the cost of defective battery piece detection in an actual scene is reduced.

Description

Technical Field

[0001] This application relates to the field of machine vision technology, and particularly to a defective solar cell detection system and method. Background Art

[0002] In the process of manufacturing solar cells, multiple process treatments can be sequentially performed on unprocessed raw materials. For example, the unprocessed raw material can be a raw silicon wafer; the raw silicon wafer is subjected to texturing treatment to obtain a textured wafer; the textured wafer is subjected to diffusion junction formation treatment to obtain a diffused wafer; the diffused wafer is subjected to coating treatment to obtain a blue film wafer. The raw silicon wafer, textured wafer, diffused wafer, blue film wafer, etc. can be collectively referred to as solar cells. However, there may be defective solar cells among the above-mentioned solar cells, such as solar cells with perforations, solar cells with hidden cracks, etc.

[0003] In order to identify defective solar cells, technicians can manually detect whether each solar cell is a defective one. However, the efficiency of manually detecting solar cells is not high. Summary of the Invention

[0004] The purpose of the embodiments of this application is to provide a defective solar cell detection system and method, which can improve the efficiency of defective solar cell detection and reduce the cost of defective solar cell detection in actual scenarios. The specific technical solutions are as follows:

[0005] In the first aspect of the embodiments of this application, a defective solar cell detection system is provided. The system includes: a conveyor belt, a laser, a camera, and a processing unit; the laser is located on one side of the detection area in the conveyor belt, and the camera is located on the other side of the detection area; where: the conveyor belt is used to convey the solar cells to be detected on the conveyor belt; the laser is used to emit laser light to the detection area when the solar cells to be detected are conveyed to the detection area; where, for the laser light emitted by the laser, the difference in light transmittance between any two different solar cells is less than a preset difference threshold; the camera is used to collect an image of the detection area under the irradiation of the laser light emitted by the laser when the solar cells to be detected are conveyed to the detection area, to obtain a to-be-detected image; and send the to-be-detected image to the processing unit; the processing unit is used to obtain the to-be-detected image; detect the solar cells to be detected in the to-be-detected image, and obtain a detection result on whether the solar cells to be detected are defective.

[0006] Optionally, the camera is a line array camera, and the laser is a line laser. The laser is specifically configured to emit laser light to the detection area at a specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area. The camera is specifically configured to collect images of the detection area under the irradiation of the laser emitted by the laser at the specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area, and splice the collected images to obtain a to-be-detected image.

[0007] Optionally, the system further includes: an encoder; the encoder is connected to the transmission shaft of the conveyor belt; the specified frequency is the counting frequency of the encoder; the encoder is configured to send a collection instruction to the camera at the specified frequency; the camera is specifically configured to collect images of the detection area under the irradiation of the laser emitted by the laser according to the received collection instruction when the battery cell to be detected starts to enter the detection area until a preset number of images are collected, and splice the preset number of collected images to obtain a to-be-detected image; wherein, the preset number is determined based on the specified frequency, the transmission speed of the conveyor belt, and the lengths of various battery cells in the transmission direction of the conveyor belt.

[0008] Optionally, the system further includes: a photoelectric sensor; along the transmission direction, the sensing area of the photoelectric sensor is located in front of the detection area; the photoelectric sensor is configured to send an arrival signal to the camera when there is an object in the sensing area; the camera is specifically configured to collect images of the detection area under the irradiation of the laser emitted by the laser according to the received collection instruction after receiving the arrival signal until a preset number of images are collected, and splice the preset number of collected images to obtain a to-be-detected image.

[0009] Optionally, the camera is further configured to send a trigger instruction to the laser when receiving the collection instruction after receiving the arrival signal; the laser is specifically configured to emit laser light to the detection area when receiving the trigger instruction.

[0010] Optionally, the processing unit is specifically configured to input the to-be-detected image into a pre-trained defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the to-be-detected image; wherein, the pre-trained defect detection model is trained using sample images containing sample battery cells; the sample images are collected under the irradiation of the laser emitted by the laser.

[0011] Optionally, the processing unit is specifically configured to determine the gray value of the image area occupied by the battery cell to be detected in the image to be detected as the gray value to be processed; determine the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs from the corresponding relationship between the preset gray value interval set and the defect detection model set, and input the image to be detected into the determined defect detection model to obtain a detection result on whether there is a defect in the battery cell to be detected in the image to be detected; wherein each defect detection model in the corresponding relationship is trained based on images collected under the laser emitted by the laser for different types of battery cells; the gray value interval corresponding to a defect detection model represents the range of gray values of the images used to train the defect detection model.

[0012] Optionally, the gray value interval set includes three gray value intervals; the defect detection model set includes three defect detection models; the three gray value intervals and the three defect detection models are in one-to-one correspondence.

[0013] Optionally, for each type of battery cell, under the laser emitted by the laser, the light transmittance of the battery cell is not less than a preset light transmittance threshold; the exposure time of the camera is negatively correlated with the preset light transmittance threshold; in any image collected by the camera under the irradiation of the laser emitted by the laser, the gray value of the image area occupied by the battery cell belongs to a preset gray interval.

[0014] Optionally, the line width of the laser emitted by the laser is not less than 5 mm and not greater than 12 mm; the laser power of the laser emitted by the laser is not less than 25 W.

[0015] Optionally, the system further includes an alarm; the processing unit is further configured to send an alarm message to the alarm in the case that the detection result indicates that there is a defect in the battery cell to be detected; the alarm is configured to give an alarm after receiving the alarm message.

[0016] In a second aspect of the embodiments of the present application, a method for detecting defective battery cells is provided, which is applied to a defective battery cell detection system, and the system includes: a conveyor belt, a laser, a camera, and a processing unit; the laser is located on one side of the detection area in the conveyor belt, and the camera is located on the other side of the detection area; the method includes:

[0017] When the battery cell to be detected on the conveyor belt is conveyed to the detection area, the laser emits laser light to the detection area; wherein, for the laser light emitted by the laser, the difference in light transmittance between any two different battery cells is less than a preset difference threshold; when the battery cell to be detected on the conveyor belt is conveyed to the detection area, the camera captures an image of the detection area under the laser light emitted by the laser to obtain a to-be-detected image; and sends the to-be-detected image to the processing unit; the processing unit obtains the to-be-detected image; detects the battery cell to be detected in the to-be-detected image, and obtains a detection result on whether the battery cell to be detected has a defect.

[0018] Optionally, the camera is a line array camera and the laser is a line laser; when the battery cell to be detected on the conveyor belt is conveyed to the detection area, the laser emits laser light to the detection area, including: when the battery cell to be detected starts to enter the detection area, the laser emits laser light to the detection area at a specified frequency until the battery cell to be detected leaves the detection area; when the battery cell to be detected on the conveyor belt is conveyed to the detection area, the camera captures an image of the detection area under the laser light emitted by the laser to obtain a to-be-detected image, including: when the battery cell to be detected starts to enter the detection area, the camera captures an image of the detection area under the laser light emitted by the laser at the specified frequency until the battery cell to be detected leaves the detection area, and splices the captured images to obtain a to-be-detected image.

[0019] Optionally, the system further includes: an encoder; the encoder is connected to the transmission shaft of the conveyor belt; the specified frequency is the counting frequency of the encoder; when the battery cell to be detected starts to enter the detection area, the camera captures an image of the detection area under the laser light emitted by the laser at the specified frequency until the battery cell to be detected leaves the detection area, and splices the captured images to obtain a to-be-detected image, including: at the specified frequency, the encoder sends a capture instruction to the camera; when the battery cell to be detected starts to enter the detection area, the camera captures an image of the detection area under the laser light emitted by the laser according to the received capture instruction until a preset number of images are captured, and splices the preset number of captured images to obtain a to-be-detected image; wherein, the preset number is determined based on the specified frequency, the conveying speed of the conveyor belt, and the lengths of various battery cells in the conveying direction of the conveyor belt.

[0020] Optionally, the system further includes: a photoelectric sensor; along the conveying direction, the sensing area of the photoelectric sensor is located in front of the detection area; when the battery cell to be detected starts to enter the detection area, the camera collects an image of the detection area under the laser emitted by the laser according to the received acquisition instruction, including: when there is an object in the sensing area, the photoelectric sensor sends an arrival signal to the camera; after receiving the arrival signal, the camera collects an image of the detection area under the laser emitted by the laser according to the received acquisition instruction.

[0021] Optionally, after receiving the arrival signal, the method further includes: when receiving the acquisition instruction, the camera sends a trigger instruction to the laser; when receiving the trigger instruction, the laser emits laser to the detection area.

[0022] Optionally, detecting the battery cell to be detected in the image to be detected to obtain a detection result on whether there are defects in the battery cell to be detected, including: inputting the image to be detected into a pre-trained defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the image to be detected; wherein, the pre-trained defect detection model is trained using sample images including sample battery cells; the sample images are collected under the laser emitted by the laser.

[0023] Optionally, inputting the image to be detected into a pre-trained defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the image to be detected, including: determining the gray value of the image area occupied by the battery cell to be detected in the image to be detected as the gray value to be processed; from the corresponding relationship between the preset gray value interval set and the defect detection model set, the processing unit determines the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs, and inputs the image to be detected into the determined defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the image to be detected; wherein, each defect detection model in the corresponding relationship is trained based on images collected under the laser emitted by the laser for different types of battery cells; the gray value interval corresponding to a defect detection model represents the range of gray values of the images used to train the defect detection model.

[0024] Optionally, the gray value interval set includes three gray value intervals; the defect detection model set includes three defect detection models; the three gray value intervals and the three defect detection models are in one-to-one correspondence.

[0025] Optionally, for each type of solar cell, under the irradiation of the laser emitted by the laser, the light transmittance of this type of solar cell is not less than a preset light transmittance threshold; the exposure time of the camera is negatively correlated with the preset light transmittance threshold; under the irradiation of the laser emitted by the laser, in any image captured by the camera, the gray value of the image area occupied by the solar cell belongs to a preset gray value interval.

[0026] Optionally, the line width of the laser emitted by the laser is not less than 5 mm and not greater than 12 mm; the laser power emitted by the laser is not less than 25 W.

[0027] Optionally, the system further includes an alarm; after detecting the solar cell to be detected in the image to be detected and obtaining the detection result of whether the solar cell to be detected has a defect, the method further includes: when the detection result indicates that the solar cell to be detected has a defect, the processing unit sends an alarm message to the alarm; after receiving the alarm message, the alarm gives an alarm.

[0028] A defective solar cell detection system provided by an embodiment of the present application includes: a conveyor belt, a laser, a camera, and a processing unit; the laser is located on one side of the detection area in the conveyor belt, and the camera is located on the other side of the detection area; the conveyor belt is used for conveying the solar cell to be detected on the conveyor belt; the laser is used for emitting laser to the detection area when the solar cell to be detected is conveyed to the detection area; for the laser emitted by the laser, the difference in light transmittance between any two different types of solar cells is less than a preset difference threshold; the camera is used for capturing an image of the detection area under the irradiation of the laser emitted by the laser when the solar cell to be detected is conveyed to the detection area, to obtain an image to be detected; and sending the image to be detected to the processing unit; the processing unit is used for obtaining the image to be detected; detecting the solar cell to be detected in the image to be detected, to obtain the detection result of whether the solar cell to be detected has a defect.

[0029] Based on the above processing, when the conveyor belt conveys the solar cell to be detected to the detection area, the laser automatically emits laser to the detection area, and the camera automatically captures an image of the detection area under the irradiation of the laser emitted by the laser, that is, captures the image to be detected of the solar cell to be detected under the irradiation of the laser emitted by the laser. Subsequently, the processing unit detects the image to be detected to obtain the detection result of whether the solar cell to be detected has a defect. That is, the defect detection of the solar cell to be detected is automatically completed by the camera, the laser, and the processing unit, and it is not necessary for technicians to manually detect whether each solar cell is a defective solar cell, which can improve the efficiency of defective solar cell detection.

[0030] Moreover, in order to effectively perform defect detection, the gray values of the images of each type of battery cell need to be within the effective detection range under the irradiation of the laser. In the embodiments of the present application, for the laser emitted by the laser device, the difference in light transmittance between any two different battery cells is less than a preset difference threshold. That is, in the present application, the light transmittance of different battery cells under the irradiation of the laser emitted by the laser device is similar, which can reduce the difference in the gray values of the images of different battery cells collected by the camera under the irradiation of the laser emitted by the laser device. Furthermore, under the irradiation of the laser emitted by the laser device, the gray values of the images of each type of battery cell collected by the camera are all within the effective detection range, reducing the influence of the difference between different battery cells on the difference in the gray values of the collected images. That is, based on the defect battery cell detection system provided in the present application, only one set of system is needed, and for different types of battery cells, the camera can collect the to-be-detected images with similar gray values under the irradiation of the laser emitted by the laser device. Subsequently, the processing unit performs defect battery cell detection on the to-be-detected images with similar gray values. There is no need to separately set up multiple sets of systems according to the types of battery cells, which can reduce the cost of defect battery cell detection in the actual scenario.

[0031] Of course, implementing any product or method of the present application does not necessarily require achieving all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following-described drawings are only some embodiments of the present application, and those of ordinary skill in the art can also obtain other embodiments based on these drawings.

[0033] Figure 1 It is the first architecture diagram of the defect battery cell detection system provided by the embodiments of the present application;

[0034] Figure 2 It is the second architecture diagram of the defect battery cell detection system provided by the embodiments of the present application;

[0035] Figure 3 It is the third architecture diagram of the defect battery cell detection system provided by the embodiments of the present application;

[0036] Figure 4 It is the fourth architecture diagram of the defect battery cell detection system provided by the embodiments of the present application;

[0037] FIG. 5(a) is an image of a raw silicon wafer with perforations provided by the embodiments of the present application;

[0038] FIG. 5(b) is an image of a raw silicon wafer with hidden cracks provided by the embodiments of the present application;

[0039] Figure 5(c) is an image of a textured wafer with perforations provided by an embodiment of the present application;

[0040] Figure 5(d) is an image of a textured wafer with a hidden crack provided by an embodiment of the present application;

[0041] Figure 5(e) is an image of a blue film with perforations provided by an embodiment of the present application;

[0042] Figure 5(f) is an image of a blue film with a hidden crack provided by an embodiment of the present application;

[0043] Figure 6 is the fifth architecture diagram of the defective cell detection system provided by an embodiment of the present application;

[0044] Figure 7 is a processing flow chart of the defective cell detection system provided by an embodiment of the present application;

[0045] Figure 8 is a flow chart of a method for detecting defective cells provided by an embodiment of the present application. Detailed implementation manners

[0046] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art based on the present application belong to the scope of protection of the present application.

[0047] In the field of machine vision technology, during the process of generating cells, multiple processings can be sequentially performed on the unprocessed raw materials. For example, the unprocessed raw material can be a raw silicon wafer; during the processing from the raw silicon wafer to the deposition of a silicon nitride film, the raw silicon wafer is subjected to a texturing process to obtain a textured wafer; the textured wafer is subjected to a diffusion junction formation process to obtain a diffused wafer; the diffused wafer is subjected to a coating process to obtain a blue film. The above raw silicon wafers, textured wafers, diffused wafers, blue films, etc. can be collectively referred to as cells. However, there may be defective cells among the above cells, such as cells with perforations, cells with hidden cracks, etc. It can be understood that even if the defective cells are further processed by the above processes, it is difficult to obtain cells that meet the production requirements. These defective cells can also be called cells with structural damage. During the process of generating cells, it is necessary to identify and remove the defective cells. To identify the defective cells, technicians can manually detect whether each cell is a defective cell. However, the efficiency of manually detecting cells is not high.

[0048] To solve the above problems, an embodiment of the present application provides a defective cell detection system. Refer to Figure 1 ,Figure 1 This is the first architecture diagram of the defective cell detection system provided by the embodiments of the present application. The system 10 includes: a conveyor belt 101, a laser 102, a camera 103, and a processing unit 104; the laser 102 is located on one side of the detection area in the conveyor belt 101, and the camera 103 is located on the other side of the detection area; wherein:

[0049] The conveyor belt 101 is used to convey the cell to be detected on the conveyor belt 101.

[0050] The laser 102 is used to emit laser light to the detection area when the cell to be detected is conveyed to the detection area; wherein, for the laser light emitted by the laser 102, the difference in light transmittance between any two different cells is less than a preset difference threshold.

[0051] The camera 103 is used to collect an image of the detection area under the irradiation of the laser light emitted by the laser 102 when the cell to be detected is conveyed to the detection area, so as to obtain the image to be detected; and send the image to be detected to the processing unit 104.

[0052] The processing unit 104 is used to obtain the image to be detected; detect the cell to be detected in the image to be detected, and obtain a detection result on whether the cell to be detected has a defect.

[0053] Based on the defective cell detection system 10 provided by the embodiments of the present application, when the conveyor belt 101 conveys the cell to be detected to the detection area, the laser 102 automatically emits laser light to the detection area, and the camera 103 automatically collects an image of the detection area under the irradiation of the laser light emitted by the laser 102, that is, collects the image to be detected of the cell to be detected under the irradiation of the laser light emitted by the laser 102. Subsequently, the processing unit 104 detects the image to be detected to obtain a detection result on whether the cell to be detected has a defect. That is, the defect detection of the cell to be detected is automatically completed by the camera 103, the laser 102, and the processing unit 104, without the need for technicians to manually detect whether each cell is a defective cell, which can improve the efficiency of defective cell detection.

[0054] Moreover, in order to effectively perform defect detection, the gray values of the images of each type of battery cell need to be within an effective detection range under the irradiation of the laser. In the embodiments of the present application, for the laser emitted by the laser 102, the difference in the light transmittance between any two different battery cells is less than a preset difference threshold. That is, in the present application, the light transmittance of different battery cells is similar under the irradiation of the laser emitted by the laser 102, which can reduce the difference in the gray values of the images of different battery cells collected by the camera 103 under the irradiation of the laser emitted by the laser 102. Furthermore, under the irradiation of the laser emitted by the laser 102, the gray values of the images of each type of battery cell collected by the camera 103 are within an effective detection range, reducing the influence of the difference between different battery cells on the difference in the gray values of the collected images. That is, based on the defective battery cell detection system 10 provided in the present application, only one set of system is needed, and for different types of battery cells, the camera 103 can collect the to-be-detected images with similar gray values under the irradiation of the laser emitted by the laser 102. Subsequently, the processing unit 104 performs defective battery cell detection on the to-be-detected images with similar gray values. There is no need to separately set up multiple sets of systems according to the types of battery cells, which can reduce the cost of defective battery cell detection in the actual scenario. It should be noted that various parameters of the laser emitted by the laser 102 (such as width, intensity, etc.) are obtained through experiments. By emitting a laser with preset parameters, the effect of the difference in the gray values of the images of different battery cells collected by the camera 103 under the laser irradiation can be achieved. In actual detection, it is not necessary to determine whether the difference in the gray values of the images of different battery cells collected by the camera 103 under the laser irradiation is really less than the preset difference threshold. Of course, in some ways, the above judgment can also be made, and the parameters of the laser can be fine-tuned during the detection process to achieve the aforementioned effect and ensure the detection effect.

[0055] Regarding the conveyor belt 101, the conveyor belt 101 can be conveyed in a preset conveying direction. In the actual scenario, the to-be-detected battery cells on the conveyor belt 101 can be placed by technicians, robots, etc. Or, after obtaining the battery cells through any of the foregoing processes, the processed battery cells can be conveyed to the conveyor belt 101 using an assembly line. The present application does not limit how the to-be-detected battery cells are on the conveyor belt 101.

[0056] The to-be-detected battery cells can be each battery cell involved in the foregoing multiple processes (hereinafter referred to as alternative battery cells); or, the alternative battery cells can also be initially inspected, and then the alternative battery cells indicated by the initial inspection results as possibly defective can be used as the to-be-detected battery cells. Exemplarily, the alternative battery cells can be initially inspected based on a machine vision method. Obviously, the to-be-detected battery cells can include various battery cells, such as the foregoing original silicon wafers, textured wafers, blue film wafers, etc.

[0057] It can be understood that there is a detection area on the conveyor belt 101 in the present application. The laser 102 is located on one side of the detection area, while the camera 103 is located on the other side of the detection area, so that the laser emitted by the laser 102 can penetrate the battery cell to be detected, and subsequently the camera 103 can collect the image of the detection area under the irradiation of the laser emitted by the laser 102.

[0058] Exemplarily, refer to Figure 2 , Figure 2 which is the second architecture diagram of the defective battery cell detection system provided by the embodiment of the present application. As Figure 2 shown, Figure 2 the battery cell 201 placed on the conveyor belt 101 represents the battery cell to be detected; the area 202 represents the detection area. The laser 102 is located below the detection area, and the laser light source in the laser 102 emits laser light towards the detection area. When the battery cell 201 is conveyed to the area 202, that is, when the battery cell to be detected is conveyed to the detection area, the laser emitted by the laser 102 can irradiate the battery cell to be detected and penetrate the battery cell to be detected.

[0059] It can be understood that if the distance between the laser emission point in the laser 102 and the battery cell to be detected is too far, the laser may not be able to penetrate the battery cell to be detected, and subsequently it will be difficult for the camera 103 to collect the image of the battery cell under the irradiation of the laser; if the distance between the laser emission point and the battery cell to be detected is too close, the laser may damage the battery cell to be detected. Therefore, in the actual scenario, after the battery cell to be detected is conveyed to the detection area, the distance between the laser emission point and the lower part of the battery cell to be detected can be limited within a certain range. The experimental results show that when the distance between the laser emission point and the lower part of the battery cell to be detected is set to 5 millimeters, the laser can penetrate the battery cell to be detected and the probability of damaging the battery cell to be detected is relatively low. Therefore, usually the distance between the laser emission point and the lower part of the battery cell to be detected can be set to 5 millimeters.

[0060] The camera 103 is located above the detection area. By adjusting the position of the lens of the camera 103, the detection area can be made to be within the field of view of the camera 103. It can be understood that if the distance between the camera 103 and the detection area is too far, the clarity of the image of the detection area collected by the camera 103 will be relatively low, and the accuracy of the detection result obtained based on the image with relatively low clarity will be relatively low; if the distance between the camera 103 and the detection area is too close, the field of view of the camera 103 may not be able to cover the detection area, that is, the camera 103 may collect a detection image including an incomplete battery cell to be detected, and then misdetection may occur when detecting defective battery cells based on the detection image. Therefore, in the actual scenario, the distance between the camera 103 and the detection area can also be limited within a certain range. The experimental results show that when the distance between the lens of the camera 103 and the detection area is not less than 275 millimeters and not more than 325 millimeters, the probability of collecting an image with relatively low clarity is small, and at this time the field of view of the camera 103 can cover the detection area. Therefore, generally, the distance between the lens of the camera 103 and the detection area can be set to be not less than 275 millimeters and not more than 325 millimeters.

[0061] For the laser 102 and the camera 103, the laser emitted by the laser 102 is an infrared laser. Since the battery cell to be detected in this application is a raw silicon wafer, or a battery cell obtained by performing a process on the raw silicon wafer, obviously, the battery cells to be detected are all silicon materials. The crystal bond structure in the silicon material causes the silicon material to be able to reflect infrared rays with a wavelength range belonging to 700 nanometers - 2500 nanometers. However, when there are defects in the battery cell to be detected, such as perforations, the crystal bond structure at the perforated position is damaged, and then the infrared rays at the perforated position will not be reflected. That is, if there are defects in the battery cell to be detected, in the image of the detection area collected by the camera 103 under the irradiation of the laser emitted by the laser 102, the gray value at the defective position on the battery cell to be detected is different from the gray values at other positions on the battery cell to be detected.

[0062] Under the irradiation of the laser emitted by the laser device 102, for any type of solar cell, such as the solar cell can be the aforementioned original silicon wafer, texturing wafer, diffusion wafer, blue film wafer, etc., the laser can penetrate through this type of solar cell. The light transmittance of this type of solar cell is the ratio of the power of the transmitted light to the total power of the laser emitted by the laser device 102 after the laser penetrates through this type of solar cell. The light transmittance of a type of solar cell is related to the surface structure, thickness, etc. of this type of solar cell. During the processing of the solar cell, the surface structure, thickness, etc. of different types of solar cells may be different, that is, the light transmittances of different types of solar cells may be different. Under the irradiation of the laser emitted by the laser device 102, the difference in the light transmittances of any two different solar cells is less than a preset difference threshold. For example, the difference in the light transmittances of the original silicon wafer and the blue film wafer is less than the preset difference threshold. In an actual scenario, the difference in the gray values of the images of any two different solar cells collected by the camera 103 can be used to represent the difference in the light transmittances of these two different solar cells. The difference in the light transmittances of any two different solar cells is less than the preset difference threshold, that is, the difference in the gray values of the images of these two solar cells collected by the camera 103 is less than the gray value difference threshold. For example, the gray value difference threshold can be 150.

[0063] Before the defective solar cell detection system 10 detects whether a solar cell has a defect, a technician can pre-configure the parameters of the laser emitted by the laser device 102 (such as the line width, laser power, etc. of the laser) to test the parameters of the laser that make the difference in the light transmittances of any two different solar cells less than the preset difference threshold, that is, to test the parameters of the laser that make the difference in the gray values of the images of any two different solar cells collected by the camera 103 less than the gray value difference threshold. The line width of the laser is the width of the aforementioned laser; the laser power is the intensity of the aforementioned laser.

[0064] For example, the technician finds through testing that when the line width of the laser is not less than 5 mm and not more than 12 mm, and the laser power is not less than 25 W, the difference in the gray values of the images of any two different solar cells collected by the camera 103 can be made less than the gray value difference threshold. Furthermore, the technician can configure the laser device 102 according to the above test results. For example, configure the line width of the laser emitted by the laser device 102 to be 6 mm and the laser power to be 30 W. Then, when the defective solar cell detection system 10 provided by the present invention detects whether a solar cell has a defect, the laser configured in advance is directly emitted by the laser device 102 to the detection area, and the difference in the light transmittances of any two different solar cells can be made less than the preset difference threshold.

[0065] It can be understood that the above limitation of "for the laser emitted by the laser, the difference in light transmittance between any two different solar cells is less than a preset difference threshold" is a limitation on the laser emitted by the laser 102 in the form of an effect, rather than a limitation on the actions (i.e., execution steps) during the actual detection process of the solar cells. That is to say, during the actual detection of the solar cells, it is not necessary to calculate the difference in light transmittance between any two different solar cells, nor to determine whether the difference is less than the preset difference threshold.

[0066] It can be understood that the laser 102 in the present application is used to emit laser light to irradiate the solar cell to be detected. Then, when the camera 103 collects an image of the detection area, the laser 102 can emit laser light to the detection area. Obviously, the laser 102 can be a surface laser or a line laser; the laser 102 can continuously emit laser light to the detection area, or it can also emit laser light to the detection area after receiving a start signal. The emission of laser light by the laser 102 to the detection area can be referred to as the laser 102 being lit.

[0067] In the case where the laser 102 is a surface laser, when the solar cell to be detected is conveyed to the detection area, the surface laser emits laser light to the detection area, and then all positions of the solar cell to be detected are under the irradiation of the laser light. At this time, the camera 103 can collect an image of the detection area to obtain a to-be-detected image including the solar cell to be detected.

[0068] Since the conveyor belt 101 moves in a preset conveying direction, that is, the position of the solar cell to be detected on the conveyor belt 101 will also change. Therefore, in order to collect an image of the detection solar cell whose position will change, the camera 103 can be a line array camera. Correspondingly, the laser 102 can also be a line laser. At this time, the method for the camera 103 to collect the to-be-detected image can refer to the detailed introduction in the subsequent embodiments. The line array camera can also be referred to as a line scan camera.

[0069] In one implementation, adjusting the field of view range of the camera 103 can enable the field of view range of the camera 103 to cover the solar cell to be detected in the detection area. If the length of the solar cell to be detected in the conveying direction is called the length of the solar cell to be detected, then the length of the solar cell to be detected in the target direction can be called the width of the solar cell to be detected. The target direction is the direction perpendicular to the conveying direction on the horizontal plane. That is, the length of the area covered by the field of view range of the camera 103 in the target direction is not less than the width of the solar cell to be detected. Therefore, according to the width of the solar cell to be detected in the actual scenario, the field of view range of the camera 103 can be set, that is, the length of the image collected by the camera 103 in the target direction (i.e., the row height in the subsequent embodiments) can be set.

[0070] After the camera 103 collects the to-be-detected image, it can send the collected image to the processing unit 104.

[0071] In one implementation, the processing unit 104 can be integrated inside the camera 103, such as the CPU (Central Processing Unit) in the camera. After the image sensor in the camera 103 captures the image to be detected, it can send the image to be detected to the CPU in the camera 103. Subsequently, the CPU detects whether there are defects in the battery cell to be detected based on the image to be detected. At this time, the camera 103 is the intelligent camera.

[0072] In another implementation, the processing unit 104 can be other devices independent of the camera 103, such as a cloud server or an industrial control computer, etc. This application does not limit this. The camera 103 can send the image to be detected to the processing unit 104 through device - to - device communication. Exemplarily, the processing unit 104 can be an industrial control computer. The camera 103 is connected to the network interface card of the industrial control computer through a network cable. After the camera 103 captures the image to be detected, it transmits the image to be detected to the industrial control computer through the network interface card.

[0073] The image to be detected can be a grayscale image; or it can also be a color image, and the processing unit 104 converts the image to be detected into a grayscale image. Subsequently, the processing unit 104 detects whether there are defects in the battery cell to be detected based on the grayscale image.

[0074] For the processing unit 104, after obtaining the image to be detected, the processing unit 104 can detect the battery cell to be detected shown in the image to be detected. Exemplarily, the processing unit 104 can process the image to be detected based on an edge - detection algorithm. If other edges are detected in addition to the edges (four sides) of the battery cell to be detected, it can be determined that there are defects in the battery cell to be detected. Then, the processing unit 104 can output the detection result. For example, when the processing unit 104 is the aforementioned industrial control computer, it can output a prompt message indicating whether there are defects in the battery cell to be detected on the display of the industrial control computer.

[0075] In some embodiments, the camera 103 is a line - array camera, and the laser 102 is a line laser; the laser 102 is specifically configured to emit laser light to the detection area at a specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area; the camera 103 is specifically configured to collect images of the detection area under the laser light emitted by the laser 102 at a specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area, and splice the collected images to obtain the image to be detected.

[0076] Since the conveyor belt 101 conveys the battery cells to be detected, that is, the positions of the battery cells to be detected change; and in the actual scenario, when starting to detect defective battery cells, the positions of the laser 102 and the camera 103 usually do not change anymore. Therefore, in order to facilitate the acquisition of images of the battery cells to be detected conveyed by the conveyor belt 101, the laser 102 can be a line laser, and the camera 103 can be a line array camera. At this time, each time the laser 102 emits laser light into the detection area, it can only transmit a part of the area of the battery cell to be detected, and the camera 103 also only acquires images of the area of the battery cell to be detected that is transmitted each time. Obviously, the laser 102 emits laser light into the detection area multiple times, and correspondingly, the camera 103 also acquires images of the detection area irradiated by the laser light emitted by the laser 102 multiple times. In this way, the camera 103 can acquire images of each area of the battery cell to be detected that is transmitted; the images of each area are stitched together, such as stitching them in the order of image acquisition, to obtain a to-be-detected image including the complete battery cell to be detected.

[0077] It can be understood that each time the laser 102 emits laser light into the detection area, the camera 103 also performs an image acquisition of the detection area. That is, the frequency at which the laser 102 emits laser light into the detection area is the same as the frequency at which the camera 103 acquires images of the detection area, both being the specified frequency. Exemplarily, the specified frequency can be calculated according to the conveying speed of the conveyor belt and the length of the area of the battery cell to be detected indicated by the image acquired by the camera 103 each time in the conveying direction. For example, if the conveying speed is 1 m / s, the length of the battery cell to be detected in the conveying direction is 210 mm, and the image acquired by the camera 103 each time is about 0.05 pix (pixels), that is, the length of the area indicated by the image acquired by the camera 103 each time in the conveying direction is about 0.05 mm, then the specified frequency is about 18000 times per second. And the number of pixel points included in the image acquired by the camera 103 each time is the same, and the subsequent acquired images can be stitched together. For example, the image acquired by the camera 103 each time can include 4096 pixel points, that is, the to-be-detected image obtained subsequently includes 4096 pixel points in the target direction.

[0078] Based on the above processing, the laser 102 emits laser light into the detection area at the specified frequency, that is, the laser 102 is lit in a stroboscopic triggering manner. Compared with a constantly lit laser, the service life of the laser 102 can be increased, and the safety hazard when using the laser 102 can be reduced.

[0079] In some embodiments, the number of times the camera 103 takes pictures can be determined according to the length of the battery cell in the conveying direction of the conveyor belt and the length of the area indicated by each image collected by the camera 103 in the conveying direction, so as to obtain a to-be-detected image containing a complete battery cell to be detected. Furthermore, after the camera 103 takes pictures the calculated number of times at the specified frequency, the camera 103 can be set to stop taking pictures.

[0080] In some embodiments, based on Figure 1 , referring to Figure 3 , Figure 3 FIG. is the third architecture diagram of the defective battery cell detection system provided by the embodiment of the present application. The defective battery cell detection system 10 further includes: an encoder 105; the encoder 105 is connected to the transmission shaft of the conveyor belt 101; the specified frequency is the counting frequency of the encoder 105;

[0081] The encoder 105 is configured to send a capture instruction to the camera 103 at a specified frequency;

[0082] The camera 103 is specifically configured to, when the battery cell to be detected starts to enter the detection area, collect an image of the detection area under the laser irradiation emitted by the laser 102 according to the received capture instruction until a preset number of images are collected, and splice the preset number of collected images to obtain a to-be-detected image. Wherein, the preset number is determined based on the specified frequency, the conveying speed of the conveyor belt 101, and the lengths of various battery cells in the conveying direction of the conveyor belt 101.

[0083] The encoder 105 can be an incremental encoder. For example, as Figure 2 shown, the encoder 105 is connected to the transmission shaft of the conveyor belt 101 ( Figure 2 not shown in the figure). In an actual scenario, the encoder 105 can be connected to the transmission shaft of the conveyor belt 101 through a fixed bracket and a coupling. When the transmission shaft rotates, it can drive the conveyor belt 101 to convey and drive the encoder 105 to count. Obviously, when the encoder 105 makes one count, that is, the transmission shaft drives the conveyor belt 101 to convey a corresponding distance. The counting frequency of the encoder 105 can indicate the conveying speed of the conveyor belt 101, that is, the encoder 105 can match the moving speed of the battery cell to be detected. For example, if the counting frequency of the encoder 105 is 18,000 times per second and the conveying speed of the conveyor belt 101 is 1 meter per second. Then when the encoder 105 makes one count, it indicates that the conveyor belt 101 has conveyed 0.05 millimeters; when the encoder 105 makes 18,000 counts, it can indicate that the conveyor belt 101 has conveyed 1 meter. Therefore, the counting frequency of the encoder 105 can be used as the specified frequency. That is, the encoder 105 can send a capture instruction to the camera 103 when counting, that is, send a capture instruction to the camera 103 at the specified frequency.

[0084] Correspondingly, when the battery cell to be detected starts to enter the detection area, the camera 103 can collect an image of the detection area under the laser irradiation emitted by the laser 102 according to the received acquisition instruction, that is, collect an image of the detection area under the laser irradiation emitted by the laser 102 at a specified frequency.

[0085] The preset number can be determined based on the specified frequency, the conveying speed of the conveyor belt 101, and the lengths of various battery cells in the conveying direction of the conveyor belt 101. Since the defective battery cell detection system 10 provided in this application can be used to detect whether various battery cells are defective, that is, for each type of battery cell, the camera 103 needs to be able to collect a to-be-detected image including the complete battery cell of this type. Therefore, the conveying distance of the conveyor belt 101 indicated by the preset number needs to be not less than the maximum value of the lengths of various battery cells in the conveying direction (hereinafter referred to as the maximum length). Exemplarily, if the maximum length is 25 cm, based on the foregoing example, the preset number can be 5000. Alternatively, in order to increase the probability that the to-be-detected image collected by the camera 103 includes the complete battery cell to be detected, the sum value of the maximum length and the preset length can be calculated as the conveying distance indicated by the preset number. If the maximum length is 25 cm and the preset length is 5 cm, based on the foregoing example, the preset number can be 6000.

[0086] Furthermore, when the battery cell to be detected starts to enter the detection area, after the camera 103 receives a preset number of acquisition instructions sent by the encoder 105, that is, after collecting a preset number of images, the camera 103 can splice the preset number of collected images to obtain the to-be-detected image.

[0087] Based on the above processing, the encoder 105 controls the camera 103 to capture images, so that the camera 103 can capture images at a specified frequency. In a scenario where the position of the battery cell to be detected changes, the resolution of the to-be-detected image captured by the camera 103 is higher, and the accuracy of the detection result obtained based on the to-be-detected image with a higher resolution is higher.

[0088] In some embodiments, the counting frequency of the encoder 105 can be set to change with the rotational speed of the transmission shaft. That is, every time the transmission shaft rotates by a specified angle, the encoder 105 makes a count. Obviously, when the rotational speed of the transmission shaft changes, the angle that the transmission shaft rotates per second changes, so the number of specified angles that the transmission shaft rotates per second changes. Furthermore, the number of times the encoder 105 makes a count per second changes, that is, the counting frequency of the encoder 105 changes. Therefore, when the conveying speed of the conveyor belt 101 changes, that is, when the rotational speed of the transmission shaft changes, the counting frequency of the encoder 105 changes accordingly, and then the image acquisition frequency of the camera 103 also changes correspondingly. In this way, the image acquisition frequency of the camera 103 can be adjusted correspondingly with the change in the conveying speed of the conveyor belt, so that the camera 103 can acquire a to-be-detected image containing a complete to-be-detected battery cell, reducing the probability that the accuracy of detecting defective battery cells is not high due to incomplete images in the subsequent process.

[0089] In some embodiments, when the camera 103 does not receive the acquisition instruction sent by the encoder 105; or, when the camera 103 receives the acquisition instruction, but the to-be-detected battery cell has not entered the detection area, the camera 103 may not acquire the image of the detection area. For example, the camera 103 can be in a standby state.

[0090] In some embodiments, on the basis of Figure 1 referring to Figure 4 Figure 4 is the fourth architecture diagram of the defective battery cell detection system provided by the embodiments of the present application. The defective battery cell detection system 10 further includes a photoelectric sensor 106; along the conveying direction, the sensing area of the photoelectric sensor 106 is located in front of the detection area; the photoelectric sensor 106 is configured to send an arrival signal to the camera 103 when there is an object in the sensing area; the camera 103 is specifically configured to, after receiving the arrival signal, acquire an image of the detection area under the laser irradiation of the laser 102 according to the received acquisition instruction until a preset number of images are acquired, and splice the preset number of acquired images to obtain a to-be-detected image.

[0091] Exemplarily, as Figure 2 shown, if the conveying direction of the conveyor belt 101 is from left to right, the sensing area of the photoelectric sensor 106 can be located on the left side of the detection area. Then, when the conveyor belt 101 conveys the to-be-detected battery cell in the conveying direction, the to-be-detected battery cell will first be conveyed to the sensing area of the photoelectric sensor 106 and then to the detection area.

[0092] When there is an object in the sensing area of the optoelectronic sensor 106, the level of the optoelectronic sensor 106 will change, that is, the optoelectronic sensor 106 is triggered and sends an arrival signal to the camera 103, indicating that the battery cell to be detected has reached the detection area. Obviously, when the optoelectronic sensor 106 is triggered, it can indicate that the battery cell to be detected is transported to the sensing area of the optoelectronic sensor 106, that is, it can indicate the incoming material position. Furthermore, after receiving the arrival signal, the camera 103 can collect an image of the detection area under the laser irradiation emitted by the laser 102 according to the received acquisition instruction to obtain the image to be detected.

[0093] Based on the above processing, when the optoelectronic sensor 106 detects that there is an object in the sensing area, it then controls the camera 103 to start image acquisition, that is, after determining through the optoelectronic sensor 106 that the battery cell to be detected is transported to the detection area, the camera 103 will start image acquisition. In this way, it can avoid the situation where the camera 103 collects an image of the detection area when there is no battery cell to be detected in the detection area, and improve the effectiveness of the image collected by the camera 103.

[0094] In some embodiments, the encoder 105 can send a trigger instruction to the laser 102 at a specified frequency. Each time the laser 102 receives the trigger instruction, it emits laser light to the detection area, that is, the laser 102 can flash and light up. Correspondingly, when the laser 102 does not receive the trigger instruction, the laser 102 can be in the off state, that is, it does not emit laser light to the detection area.

[0095] In some embodiments, the camera 103 is further configured to send a trigger instruction to the laser 102 when receiving the acquisition instruction after receiving the arrival signal; the laser 102 is specifically configured to emit laser light to the detection area when receiving the trigger instruction.

[0096] When the camera 103 receives the acquisition instruction but does not receive the arrival signal, that is, when the battery cell to be detected has not been transported to the detection area, the camera 103 does not need to send a trigger instruction to the laser 102 when receiving the acquisition instruction.

[0097] After the camera 103 receives the arrival signal and when receiving the acquisition instruction, the camera 103 can send a trigger instruction to the laser 102 so that when the laser 102 receives the trigger instruction, it emits laser light to the detection area. And obviously, after receiving the arrival signal, when the camera 103 sends a preset number of trigger instructions to the laser 102, that is, the camera 103 has collected a preset number of images. At this time, even if the camera 103 receives the acquisition instruction again, the camera 103 does not need to send a trigger instruction to the laser 102 anymore. Correspondingly, when the laser 102 does not receive the trigger instruction, the laser 102 can be in the off state, that is, it does not emit laser light to the detection area.

[0098] Based on the above processing, when the camera 103 is about to capture an image, the camera 103 sends a trigger signal to the laser 102 to trigger the laser 102 to emit laser light towards the detection area. In this way, it can be ensured that the detection area is irradiated with laser light when taking pictures, which can improve the effectiveness of the laser 102 in emitting laser light.

[0099] In some embodiments, in order to enable the camera 103 to capture an image under the irradiation of the laser light emitted by the laser 102, the duration for which the laser 102 remains lit each time it receives a trigger command may not be less than the duration for the camera 103 to capture one image. For example, when the duration required for the camera 103 to capture one image is 20 milliseconds, the duration for the laser 102 to emit laser light after receiving the trigger command may be 50 milliseconds.

[0100] In some embodiments, for each type of solar cell, under the irradiation of the laser light emitted by the laser 102, the light transmittance of this type of solar cell is not less than a preset light transmittance threshold; the exposure time of the camera 103 is negatively correlated with the preset light transmittance threshold; in any image captured by the camera 103 under the irradiation of the laser light emitted by the laser 102, the gray value of the image area occupied by the solar cell belongs to a preset gray value interval.

[0101] As mentioned above, the light transmittances of different solar cells are different. In order to make the difference between the light transmittances of any two different solar cells smaller, the line width of the laser light emitted by the laser 102 can be increased. Since the line width of the laser is proportional to the spot diameter of the laser, after the line width of the laser increases, the spot of the laser also increases, resulting in a decrease in the parallelism and density of the laser. Since for any type of solar cell, there is a logarithmic relationship between the density of the laser and the light transmittance of this solar cell under the irradiation of this laser, when the original light transmittance of this type of solar cell is relatively large, after the density of the laser decreases, the light transmittance of this type of solar cell decreases more; when the original light transmittance of this type of solar cell is relatively small, after the density of the laser decreases, the light transmittance of this type of solar cell decreases less. Then, after increasing the line width of the laser, the difference between the light transmittances of different types of solar cells decreases, making the difference between the light transmittances of any two different solar cells smaller. Then, under the irradiation of the laser light emitted by the laser 102, the light transmittance of each type of solar cell is not less than the preset light transmittance threshold, that is, the light transmittances of various solar cells are not too small.

[0102] It can be understood that under the irradiation of the laser emitted by the laser 102, the lower the laser power, the smaller the gray value of the image area occupied by the battery cell in the image collected by the camera 103 (hereinafter simply referred to as the gray value of the image); the higher the laser power, the larger the gray value of the image area occupied by the battery cell in the image collected by the camera 103. In order to avoid the image finally collected by the camera 103 being too dark, that is, in order to avoid the gray value of the image area occupied by the battery cell in the collected image being too small, by adjusting the laser power, the gray value of the image of the battery cell collected by the camera 103 under the irradiation of the laser emitted by the laser 102 is not less than a preset gray value threshold. Exemplarily, the laser power can be 25 watts, and the preset gray value threshold can be 70.

[0103] In order to avoid the image finally collected by the camera 103 being too bright, that is, in order to avoid the gray value of the image area occupied by the battery cell in the collected image being too large, the exposure time of the camera 103 can be adjusted so that in any image collected by the camera 103 under the irradiation of the laser emitted by the laser 102, the gray value of the image area occupied by the battery cell belongs to a preset gray value interval. Obviously, the exposure time of the camera 103 is negatively correlated with the preset light transmittance threshold. The larger the preset light transmittance threshold, the greater the light power transmitted by any battery cell under the irradiation of the laser emitted by the laser 102, and the shorter the exposure time of the camera 103. The preset gray value interval can be determined according to the ambient brightness in the actual scene. For example, usually, the gray values of various battery cells collected under the ambient brightness in the actual scene belong to the interval [70, 240]. Then, in order to be more in line with the actual scene, such as when the usage time of the lighting equipment in the actual scene increases, the ambient brightness may change. The preset gray value interval can be determined based on this interval. For example, the preset gray value interval can be set to [100, 230], or it can also be set to [80, 210].

[0104] In some embodiments, the line width of the laser emitted by the laser 102 is not less than 5 mm and not greater than 12 mm; the laser power emitted by the laser 102 is not less than 25 watts.

[0105] Under normal circumstances, when using a laser to irradiate a battery cell to detect whether there are defects in the battery cell, the line width of the laser emitted by the laser is usually not less than 2 mm and not greater than 3 mm. As mentioned above, after increasing the line width of the laser, the difference in the light transmittance between different types of battery cells will decrease. When the line width of the laser is not less than the lower limit of the line width, the difference in the light transmittance between different types of battery cells will be less than the preset difference threshold. As mentioned above, after the line width of the laser increases, the light transmittance of the battery cell will decrease. It can be understood that after the line width of the laser reaches the upper limit of the line width, the laser cannot transmit through some battery cells, and naturally, the camera 103 cannot collect the image to be detected. Therefore, the line width of the laser is not greater than the upper limit of the line width. Therefore, in order to enable the defective battery cell detection system 10 provided by the embodiments of the present application to perform defect detection on different types of battery cells, that is, the laser emitted by the laser 102 can transmit through various battery cells, and under the irradiation of the laser emitted by the laser 102, the difference in the light transmittance between any two battery cells is less than the preset difference threshold. That is, the line width of the laser emitted by the laser 102 is not less than the lower limit of the line width and not greater than the upper limit of the line width. Exemplarily, when the upper limit of the line width is 5 mm, the lower limit of the line width is 12 mm; or, when the upper limit of the line width is 6 mm, the lower limit of the line width is 10 mm. The specific values of the upper limit and the lower limit of the line width can be determined according to the test results in the actual scenario, and the present application does not limit this.

[0106] As mentioned above, when the line width of the laser increases and the light transmittance of the battery cell decreases, if the power of the laser emitted by the laser 102 remains unchanged, it may cause the power of the light transmitted through the battery cell to be too small, and then cause the image collected by the camera 103 to be too dark. Therefore, in order to avoid the image finally collected by the camera 103 from being too dark, the power of the laser emitted by the laser 102 can be increased. Under normal circumstances, the laser power is 15 W. When the laser power is not less than the lower limit of the power, in any image collected by the camera 103 under the irradiation of the laser emitted by the laser 102, the gray value of the image area occupied by the battery cell belongs to the preset gray value interval. It can be understood that after the laser power is increased, it may cause the power of the light transmitted through the battery cell to be too large, and then the image collected by the camera 103 may be too bright. As mentioned above, at this time, the exposure time of the camera 103 can be adjusted to avoid the image finally collected by the camera 103 from being too bright. When the laser power is increased to the upper limit of the power, even if the exposure time of the camera 103 is adjusted to the minimum, the image collected by the camera 103 will still be too bright.

[0107] Therefore, in order to reduce the probability that the image captured by the camera 103 is too bright or too dark, the laser power emitted by the laser 102 is not less than the lower power limit and not greater than the upper power limit. Exemplarily, when the lower power limit is 25 watts, the upper power limit can be 60 watts; or, when the lower power limit is 30 watts, the upper power limit can be 50 watts. The specific values of the upper power limit and the lower power limit can be determined according to the test results in the actual scenario, and the present application does not limit this.

[0108] Based on the above processing, when the conveyor belt 101 conveys the battery slice to be detected to the detection area, the laser 102 automatically emits laser light to the detection area, and the camera 103 automatically captures an image of the detection area under the irradiation of the laser light emitted by the laser 102, that is, captures the image to be detected of the battery slice to be detected under the irradiation of the laser light emitted by the laser 102. Subsequently, the processing unit 104 detects the image to be detected to obtain a detection result on whether there are defects in the battery slice to be detected. That is, the defect detection of the battery slice to be detected is automatically completed through the camera 103, the laser 102, and the processing unit 104, without the need for technicians to manually detect whether each battery slice is a defective battery slice, which can improve the efficiency of defective battery slice detection. Moreover, in order to effectively perform defect detection, the gray value of the image of each type of battery slice needs to be within an effective detection range under the irradiation of the laser. In the embodiments of the present application, for the laser light emitted by the laser 102, the difference in the light transmittance of any two different battery slices is less than a preset difference threshold, that is, under the irradiation of the laser light emitted by the laser 102 in the present application, the light transmittance of different battery slices is similar, which can reduce the difference in the gray values of the images of different battery slices captured by the camera 103 under the irradiation of the laser light emitted by the laser 102. Furthermore, under the irradiation of the laser light emitted by the laser 102, the gray value of the image of each type of battery slice captured by the camera 103 is within an effective detection range, reducing the influence of the difference between different battery slices on the difference in the gray values of the captured images. That is, based on the defective battery slice detection system 10 provided by the present application, only one set of system is needed, and for different types of battery slices, the camera 103 can capture images to be detected with similar gray values under the irradiation of the laser light emitted by the laser 102. Subsequently, the processing unit 104 performs defective battery slice detection on these images to be detected with similar gray values. There is no need to separately set up multiple sets of systems according to the types of battery slices, which can reduce the cost of defective battery slice detection in the actual scenario.

[0109] In some embodiments, the processing unit 104 is specifically configured to input the image to be detected into a pre-trained defect detection model to obtain a detection result on whether there are defects in the battery slice to be detected in the image to be detected. Among them, the pre-trained defect detection model is trained using sample images containing sample battery slices; the sample images are captured under the irradiation of the laser light emitted by the laser 102.

[0110] The pre-trained defect detection model is deployed in the processing unit 104. The training device for training the defect detection model can be the processing unit 104 itself, such as the aforementioned industrial control computer; or it can also be other devices, which is not limited in this application. The defect detection model can also be referred to as a visual detection software or algorithm software that performs detection based on machine vision technology.

[0111] In one implementation, the training device can obtain a sample image containing sample solar cells and a sample label indicating whether the sample solar cells shown in the sample image are defective. At this time, the sample solar cells can include various solar cells, such as the aforementioned raw silicon wafers, textured wafers, blue film wafers, etc. The sample image is input into the defect detection model with an initial structure to obtain a prediction result indicating whether the sample solar cells shown in the sample image are defective. Then, the loss function value representing the difference between the sample label and the prediction result is calculated, and the model parameters of the defect detection model with the initial structure are adjusted based on the calculated loss function value until the preset convergence condition is reached, and the trained defect detection model is obtained. That is, a general defect detection model is jointly trained based on the sample images of various solar cells. Subsequently, after the processing unit 104 obtains the image to be detected, it can directly detect the image to be detected based on the pre-trained defect detection model to obtain the detection result indicating whether the solar cells to be detected are defective. The defect detection model can be a model constructed based on SSD (Single Shot MultiBox Detector), a model constructed based on YOLO (You Only Look Once), etc., which is not limited in this application.

[0112] In another implementation, after the training device obtains the sample image, it can first determine the sample image area occupied by the minimum bounding rectangle of the sample solar cells shown in the sample image in the sample image, and then train the defect detection model based on the sample image area. Subsequently, after the processing unit 104 obtains the image to be detected, it can first extract the image area occupied by the minimum bounding rectangle of the solar cells to be detected in the image to be detected, and then detect the image area based on the pre-trained defect detection model to obtain the detection result indicating whether the solar cells to be detected are defective.

[0113] In some embodiments, the processing unit 104 is specifically configured to determine the gray value of the image area occupied by the solar cells to be detected in the image to be detected as the gray value to be processed; determine the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs from the correspondence between the preset gray value interval set and the defect detection model set, and input the image to be detected into the determined defect detection model to obtain the detection result indicating whether the solar cells to be detected in the image to be detected are defective.

[0114] Among them, each defect detection model in the correspondence relationship is obtained by training on images collected under the laser irradiation emitted by the laser 102 for different types of solar cells; the gray value range corresponding to a defect detection model represents the range of gray values of the images used to train this defect detection model.

[0115] As described above, the original silicon wafer, the textured wafer, the diffused wafer, the blue film wafer, etc. can be collectively referred to as solar cells. And the morphological changes of solar cells caused by different process treatments are different. When the morphological change of a solar cell is large after a process treatment, the solar cell obtained through this process treatment can be called a result wafer; correspondingly, when the morphological change of a solar cell is small after a process treatment, the solar cell obtained through this process treatment can be called a process wafer. For example, the above-mentioned original silicon wafer, textured wafer, and blue film wafer are result wafers; the above-mentioned diffused wafer is a process wafer. It can be understood that the images of different types of result wafers are quite different, but the image of a process wafer is less different from the image of a type of result wafer.

[0116] Exemplarily, FIG. 5(a) is an image of an original silicon wafer with a perforation provided by an embodiment of the present application; FIG. 5(b) is an image of an original silicon wafer with a hidden crack provided by an embodiment of the present application; FIG. 5(c) is an image of a textured wafer with a perforation provided by an embodiment of the present application; FIG. 5(d) is an image of a textured wafer with a hidden crack provided by an embodiment of the present application; FIG. 5(e) is an image of a blue film wafer with a perforation provided by an embodiment of the present application; FIG. 5(f) is an image of a blue film wafer with a hidden crack provided by an embodiment of the present application;. In FIGS. 5(a), 5(c), and 5(e) above, the white dots indicate the perforations on the solar cell; in FIGS. 5(b), 5(d), and 5(f) above, the black lines indicate the hidden cracks in the solar cell. Based on the above Figure 5(a) - Figure 5(f) It can be seen that the differences between the images of the original silicon wafer, the textured wafer, and the blue film wafer are quite large.

[0117] Obviously, if a general defect detection model is trained using images of different types of solar cells, the accuracy of the detection results of the defect detection model for different types of solar cells may not be high. Therefore, in order to improve the accuracy of the detection results of whether the solar cell to be detected has defects, the training device can train the defect detection model with the initial structure respectively based on the images collected when different types of solar cells are irradiated by the laser emitted by the laser device 102, and obtain multiple trained defect detection models. That is, multiple defect detection models applicable to detecting defects in this type of solar cell can be trained based on the type of solar cell. And it can be understood that for each type of solar cell, in the image of this type of solar cell collected under the irradiation of the laser emitted by the laser device 102, the gray values of the image area occupied by this type of solar cell generally belong to the same gray value interval. And the gray value interval is related to the ambient brightness in the actual scenario. When the ambient brightness in the actual scenario is high, both the upper and lower limits of the gray value interval corresponding to this type of solar cell are large; when the ambient brightness in the actual scenario is low, both the upper and lower limits of the gray value interval corresponding to this type of solar cell are small. The gray value of an image area can be the statistical value of the gray values of each pixel point in this image area. Correspondingly, the gray value of an image area can also be called the average gray value of the solar cell imaging in the image.

[0118] In order to obtain multiple defect detection models trained based on the type of solar cell, for each type of solar cell, the training device can determine the gray value interval corresponding to this type of solar cell. Exemplarily, the training device can obtain multiple images of this type of solar cell under the irradiation of the laser emitted by the laser device 102 under the ambient brightness conditions of the actual scenario, and then synthesize the gray values of the image area occupied by this type of solar cell in the multiple images to determine the gray value interval corresponding to this type of solar cell. For example, the maximum and minimum values of the gray values of the image area occupied by this type of solar cell can be determined, and then the interval from the minimum value to the maximum value can be determined as the gray value interval corresponding to this type of solar cell. Or, in order to improve the fault tolerance rate, the interval from the minimum value to the maximum value can be expanded, and the expanded interval can be used as the gray value interval corresponding to this type of solar cell. Obviously, the method for determining the gray value interval corresponding to each type of solar cell is not limited to this, and this application does not limit this.

[0119] It can be understood that the difference between the image of the process wafer and the image of a certain type of result wafer is small. The training device can use the union of the gray value range corresponding to this type of result wafer and the gray value range corresponding to the process wafer with a small difference from the image of this type of result wafer as a gray value range (that is, the gray value range included in the preset gray value range set); correspondingly, use the image of this type of result wafer and the image of the process wafer with a small difference from the image of this type of result wafer to jointly train to obtain a defect detection model, and obtain the defect detection model corresponding to this gray value range. For example, the process wafer can be the aforementioned diffusion wafer, and the result wafer can be the aforementioned textured wafer. The training device can use the union of the gray value range corresponding to the diffusion wafer and the gray value range corresponding to the textured wafer as a gray value range; use the images of the diffusion wafer and the textured wafer to jointly train to obtain a defect detection model. Subsequently, the processing unit 104 can use this defect detection model to detect the image whose gray value to be processed belongs to this union. That is, one gray value range corresponds to one defect detection model; for each gray value range, there may be multiple types of battery wafers whose gray values of the images belong to this gray value range.

[0120] Exemplarily, as shown in FIGS. 5(a) and 5(c), under the same ambient brightness, the gray value of the image of the original silicon wafer is smaller than the gray value of the image of the textured wafer; as shown in FIGS. 5(c) and 5(e), under the same ambient brightness, the gray value of the image of the textured wafer is smaller than the gray value of the image of the blue film wafer. And as described above, the difference between the image of the diffusion wafer and the image of the textured wafer is small, that is, the gray value of the image of the diffusion wafer is close to the gray value of the image of the textured wafer.

[0121] In some embodiments, the gray value range set includes three gray value ranges; the defect detection model set includes three defect detection models; the three gray value ranges and the three defect detection models correspond one by one.

[0122] As introduced above, the aforementioned original silicon wafer, textured wafer, and blue film wafer are result wafers; the aforementioned diffusion wafer is a process wafer. The differences between the images of different types of result wafers are large, but the difference between the image of the process wafer and the image of a certain type of result wafer is small. In the actual scenario, among the images of the original silicon wafer, textured wafer, and blue film wafer collected by the camera 103, the differences in the gray values of the image areas occupied by the battery wafers are large, while in the images of the diffusion wafer and the textured wafer collected, the differences in the gray values of the image areas occupied by the battery wafers are small, that is, the gray value corresponding to the diffusion wafer is close to the gray value corresponding to the textured wafer.

[0123] The experimental results show that when the parameters of the laser emitted by the laser 102 (including the line width, laser power, etc. of the laser) fall within the ranges introduced in the foregoing embodiments, at the ambient brightness in the actual scenario, the gray values of the images captured by the camera 103 generally fall within three gray value intervals. According to the magnitudes of the numerical values at the midpoints of the three gray value intervals, the three gray value intervals can be sorted, and there may be an intersection or no intersection between two adjacent gray value intervals in sequence. For each two adjacent gray value intervals in sequence, if the two gray value intervals include the same gray value, that is, there is an intersection between the two gray value intervals; if the two gray value intervals do not include the same gray value, that is, there is no intersection between the two gray value intervals. For example, if the two gray value intervals have the same endpoint and at least one of the two gray value intervals does not include this endpoint, there is no intersection between the two gray value intervals; or, in the two gray value intervals, the upper limit of the gray value interval with the smaller corresponding gray value is less than the lower limit of the gray value interval with the larger corresponding gray value, and there is no intersection between the two gray value intervals.

[0124] Exemplarily, the preset gray value interval set includes a first gray value interval [70, 130], a second gray value interval [140, 200], and a third gray value interval [180, 240]. In the order of magnitude, the first gray value interval is adjacent to the second gray value interval. In these two gray value intervals, the upper limit of the first gray value interval with the smaller order of magnitude (i.e., 130) is less than the lower limit of the second gray value interval with the larger order of magnitude (i.e., 140). Therefore, there is no intersection between the first gray value interval and the second gray value interval. The second gray value interval is adjacent to the third gray value interval. In these two gray value intervals, the upper limit of the second gray value interval with the smaller order of magnitude (i.e., 200) is greater than the lower limit of the third gray value interval with the larger order of magnitude (i.e., 180). Therefore, there is an intersection [180, 200] between the second gray value interval and the third gray value interval.

[0125] Correspondingly, the defect detection models usually obtained by training based on the images captured when different types of solar cells are irradiated by the laser emitted by the laser 102 also generally include three, and the three defect detection models correspond one-to-one with the three gray value intervals. As in the foregoing example, the defect detection model obtained by training based on the sample image showing the original silicon wafer corresponds to the gray value interval [70, 130]; the defect detection models obtained by training based on the sample images showing the textured wafer and the diffused wafer correspond to the gray value interval [140, 200]; the defect detection model obtained by training based on the sample image showing the blue film wafer corresponds to the gray value interval [180, 240].

[0126] As described above, there may be an intersection between two adjacent gray - scale value intervals in ascending order. If the gray - scale value to be processed belongs to the intersection, the gray - scale value to be processed will belong to two gray - scale value intervals. At this time, the processing unit 104 can comprehensively consider the two gray - scale value intervals to which the gray - scale value to be processed belongs and jointly detect the image to be detected.

[0127] In one way, the processing unit 104 can calculate the differences between the gray - scale value to be processed and the mid - point values of the two gray - scale value intervals to which it belongs respectively, and then use the defect detection model corresponding to the gray - scale value interval to which the value with the smallest difference from the gray - scale value to be processed belongs to detect the image to be detected. For example, if the gray - scale value to be processed is 185, based on the above example, the difference between the gray - scale value to be processed and the mid - point value (i.e., 170) of the second gray - scale value interval [140, 200] is 15; the difference between the gray - scale value to be processed and the mid - point value (i.e., 210) of the third gray - scale value interval [180, 240] is 25. Since 25 is greater than 15, the processing unit 104 can use the defect detection model corresponding to the second gray - scale value interval to detect the image to be detected.

[0128] In another way, the processing unit 104 can use the defect detection models corresponding to the two gray - scale value intervals to which the gray - scale value to be processed belongs respectively to detect the image to be detected, and obtain two reference detection results. By synthesizing the two reference detection results, the detection result of whether there are defects in the battery cell to be detected in the image to be detected is obtained. If both reference detection results indicate that there are no defects in the battery cell to be detected in the image to be detected, then the final detection result is that there are no defects in the battery cell to be detected in the image to be detected; if there is a reference detection result indicating that there are defects in the battery cell to be detected in the image to be detected among the two reference detection results, then the final detection result indicates that there are defects in the battery cell to be detected in the image to be detected.

[0129] Based on the above example, the processing unit 104 uses the defect detection model corresponding to the second gray - scale value interval to detect the image to be detected and obtains a reference detection result indicating that there are defects in the battery cell to be detected; uses the defect detection model corresponding to the third gray - scale value interval to detect the image to be detected and obtains a reference detection result indicating that there are no defects in the battery cell to be detected. Since there is a reference detection result indicating that there are defects in the battery cell to be detected in the image to be detected among the two reference detection results, the processing unit 104 finally outputs a detection result indicating that there are defects in the battery cell to be detected in the image to be detected.

[0130] In one implementation, for each gray value interval, the training device can obtain a sample image whose gray value of the image belongs to this gray value interval, that is, obtain a sample image containing sample solar cells of the type corresponding to this gray value interval, as well as a sample label indicating whether the sample solar cells shown in the sample image are defective. Input the sample image into the defect detection model with the initial structure to obtain a prediction result indicating whether the sample solar cells shown in the sample image are defective. Then, calculate the value of the loss function representing the difference between the sample label and the prediction result, and adjust the model parameters of the defect detection model with the initial structure based on the calculated value of the loss function until the preset convergence condition is reached, and obtain the trained defect detection model, that is, obtain the defect detection model corresponding to this gray value interval. The defect detection model corresponding to a gray value interval can be dedicated to detecting whether the solar cells to be detected in the image to be detected whose gray value belongs to this gray value interval are defective.

[0131] Correspondingly, the defect detection models corresponding to each gray value interval can be deployed in the processing unit 104. After the processing unit 104 obtains the image to be detected, it can extract the image area occupied by the solar cells to be detected from the image to be detected, and then calculate the average value of the gray values of each pixel point in the extracted image area to obtain the gray value to be processed. Determine the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs. Subsequently, use the determined defect detection model to detect the image to be detected to obtain a detection result indicating whether the solar cells to be detected are defective.

[0132] In another implementation, for each gray value interval, after the training device obtains a sample image whose gray value of the image belongs to this gray value interval, it can determine the sample image area occupied by the minimum bounding rectangle of the sample solar cells shown in the sample image in the sample image, and then train the defect detection model corresponding to this gray value interval based on the sample image area. Subsequently, after the processing unit 104 determines the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs, it can use the determined defect detection model to detect the image area occupied by the minimum bounding rectangle of the solar cells to be detected in the image to be detected to obtain a detection result indicating whether the solar cells to be detected are defective.

[0133] Based on the above processing, the training device can respectively train defect detection models corresponding to each gray value interval based on the images collected when different types of solar cells are irradiated by the laser emitted by the laser 102. Subsequently, according to the to-be-processed gray value corresponding to the to-be-detected solar cell in the to-be-detected image, the defect detection model corresponding to the gray value interval to which the to-be-processed gray value belongs is used to detect the to-be-detected solar cell in the to-be-detected image. That is, using the defect detection model applicable to process the images in this gray value interval to process the images whose gray values belong to this gray value interval can improve the accuracy of defect detection for the to-be-detected solar cell.

[0134] In some embodiments, on the basis of Figure 1 participate in Figure 6 , Figure 6 is the fifth architecture diagram of the defective solar cell detection system provided by the embodiments of the present application. The defective solar cell detection system 10 further includes an alarm 107; a processing unit 104, which is further configured to send an alarm message to the alarm 107 when the detection result indicates that the to-be-detected solar cell has a defect; the alarm 107 is configured to give an alarm after receiving the alarm message.

[0135] Exemplarily, as Figure 2 shown, after the camera 103 collects the to-be-detected image, it sends the to-be-detected image to the processing unit 104; after the processing unit 104 obtains the detection result indicating that the to-be-detected solar cell has a defect, it can display the detection result on the display 204; or it can send an alarm message to the control terminal in the alarm 107, such as a PLC (Programmable Logic Controller, programmable controller device) or a motion control board. After receiving the alarm message, the alarm 107 can give an alarm, and the specific form of the alarm can be set according to the actual scenario. For example, the alarm 107 can emit an indicating sound and control the alarm light 203 to flash a specified color of light, etc., to prompt that a defective solar cell has been detected. Subsequently, the technician can remove the defective solar cell from the conveyor belt, that is, perform the processing work of relevant defective products. Or, the alarm 107 can also send a signal to the removal device to instruct the removal device to remove the defective solar cell from the conveyor belt, and the present application does not limit this.

[0136] Based on the above processing, after the processing unit 104 detects a defective solar cell, an alarm can be given through the alarm 107. Subsequently, based on the alarm of the alarm 107, corresponding processing can be performed on the defective solar cell. That is, corresponding processing can be timely performed on the defective solar cell, reducing the amount of useless processing and improving the effectiveness of processing.

[0137] In some embodiments, when a machine vision-based method is used to conduct a preliminary inspection on alternative solar cells, among the alternative solar cells with preliminary inspection results indicating possible defects, there may be solar cells with structural damage (hereinafter referred to as the solar cells to be rejected), and solar cells with surface damage (hereinafter referred to as the repairable solar cells). For example, the solar cells to be rejected are solar cells with perforations or hidden cracks; the repairable solar cells are solar cells with poor coating or oil stains on the surface. Obviously, the crystal bond structure in the solar cells to be rejected is damaged, while the crystal bond structure in the repairable solar cells is not damaged.

[0138] Then, in the case where the solar cells to be detected in this application are alternative solar cells with possible defects preliminarily detected by a machine vision-based method, the solar cells to be detected may include the above-mentioned solar cells to be rejected and repairable solar cells.

[0139] And as mentioned above, when the crystal bond structure in the silicon material is damaged, in the image of the detection area under the laser irradiation emitted by the laser 102 collected by the camera 103, the gray value at the defective position on the solar cell to be detected is different from the gray value at other positions on the solar cell to be detected. It can be seen that based on the defective solar cell detection system 10 provided in this application, the detected defective solar cells are the solar cells with damaged crystal bond structure, that is, the detected ones are the solar cells to be rejected.

[0140] It can be understood that there are already structural damages in the solar cells to be rejected. Even if the solar cells to be rejected are further processed, it is difficult to obtain normal solar cells that meet the production requirements. However, the crystal bond structure in the repairable solar cells is not damaged, and there are only surface damages. After the repairable solar cells are reprocessed, normal solar cells that meet the production requirements can still be obtained; for example, by re-coating the solar cells with poor coating, blue-coated wafers can still be obtained; by re-texturing the solar cells with oil stains on the surface, textured wafers can still be obtained.

[0141] Therefore, based on the defective solar cell detection system 10 provided in this application, the solar cells to be rejected can be detected from the solar cells to be detected. Correspondingly, the solar cells to be rejected can be removed subsequently, avoiding useless processing of the solar cells to be rejected and improving the effectiveness of the processing. And obviously, the solar cells to be detected without structural damage are the repairable solar cells that can still obtain normal solar cells through processing. Subsequently, the repairable solar cells can be further processed accordingly, reducing the loss in the production process of solar cells.

[0142] In some embodiments, refer to Figure 7 。 Figure 7 It is a processing flow chart of the defective solar cell detection system provided in the embodiments of this application.

[0143] S1: Inflow of solar cells.

[0144] In this step, the inflowing solar cells are the solar cells to be detected in the foregoing embodiments, such as the solar cells that can be conveyed to the conveyor belt 101 through the assembly line. The conveyor belt 101 can convey the solar cells to be detected in accordance with a preset conveying direction. In an actual scenario, the conveyor belt 101 can be the conveyor belt on the rework machine.

[0145] S2: Determine whether the photoelectric switch is triggered.

[0146] In this step, the photoelectric switch is the photoelectric sensor 106 in the foregoing embodiments. When the conveyor belt 101 conveys the solar cells to be detected to the sensing area of the photoelectric switch, that is, when there is an object in the sensing area of the photoelectric switch, the photoelectric switch is triggered. When the photoelectric switch is not triggered, it can return to step S1 and continue to wait for the inflow of solar cells. When the photoelectric switch is triggered, it can send an arrival signal to the camera 103 and enter step S3.

[0147] S3: The camera starts to scan and output images according to the count of the encoder.

[0148] In this step, when the conveyor belt 101 is conveying, the encoder 105 can perform counting and send an acquisition instruction to the camera 103. After the photoelectric switch is triggered, that is, after the camera 103 receives the arrival signal, the camera 103 starts to send a trigger instruction to the laser 102 and acquires an image of the detection area under the laser irradiation emitted by the laser 102 according to the acquisition instruction sent by the encoder 105. The trigger instruction can be called a line trigger signal; the camera 103 acquires images at a specified frequency, which can be called acquiring images in the way of frame trigger image acquisition.

[0149] S4: The laser starts to strobe.

[0150] In this step, the laser is the laser 102 in the foregoing embodiments. Laser strobing means that the laser 102 in the foregoing embodiments emits laser to the detection area at a specified frequency. After the camera 103 receives the arrival signal, every time the encoder 105 sends an acquisition instruction to the camera 103, the camera 103 can send a trigger instruction to the laser 102; every time the laser 102 receives the trigger instruction, it can emit laser to the detection area, that is, realize laser strobing. The trigger instruction can be called a line signal.

[0151] S5: Output images when meeting the fixed line height, and reset the switch.

[0152] In this step, the fixed line height is the length of the image captured by the camera 103 in the target direction in the foregoing embodiment. Each time the camera 103 captures an image, when the length of the captured image in the target direction reaches the fixed line height, the camera 103 finishes capturing the current image. After the camera 103 captures a preset number of images, the preset number of captured images can be stitched together to obtain the image to be detected.

[0153] After the camera 103 captures a preset number of images, it is indicated that the image capture for the current battery cell to be detected has been completed, and the battery cell to be detected is conveyed out of the detection area. At this time, there is no object in the sensing area of the photoelectric switch, so the level of the photoelectric switch will return to the state when there is no object in the sensing area, that is, the photoelectric switch is not triggered, which means the switch is reset. When the next battery cell to be detected is conveyed to the sensing area of the photoelectric switch, the level of the photoelectric switch will change again, that is, the photoelectric switch is triggered again.

[0154] S6: Transmit the image to the algorithm software on the industrial control computer.

[0155] In this step, the image is the image to be detected in the foregoing embodiment, and the industrial control computer is the foregoing processing unit 104. After the camera 103 captures the image to be detected, it transmits the image to be detected to the industrial control computer through the network interface capture card.

[0156] S7: Determine whether the gray level of the battery cell is less than 100.

[0157] In this step, the gray level of the battery cell is the gray level value to be processed in the foregoing embodiment. After the processing unit 104 obtains the image to be detected, it can first extract the area of the image to be detected occupied by the minimum circumscribed rectangle of the battery cell to be detected in the image to be detected, and then calculate the average value of the gray level values of each pixel point in the area of the image to be detected as the gray level of the battery cell. Then determine whether the gray level of the battery cell is less than 100. If the gray level of the battery cell is less than 100, step S8 is executed; if the gray level of the battery cell is not less than 100, step S9 is executed.

[0158] S8: Use the deep learning example with a gray level of 100 to detect the image.

[0159] In this step, the deep learning demo (example) with a gray level of 100 is the defect detection model corresponding to the gray level range [70, 130] in the foregoing embodiment. When the gray level of the battery cell is less than 100, the processing unit 104 can use the deep learning demo with a gray level of 100 to detect the image to be detected to obtain the detection result, that is, complete the detection and analysis of the quality of the battery cell based on the image processing algorithm corresponding to the gray level range [70, 130]; then enter step S12.

[0160] S9: Determine whether the gray level of the battery cell is less than 200.

[0161] In this step, when the gray level of the cell is not less than 100, the processing unit 104 can continue to determine whether the gray level of the cell is less than 200. When the gray level of the cell is less than 200, step S10 is executed; when the gray level of the cell is not less than 200, step S11 is executed.

[0162] S10: Detect the image using the deep learning example with a gray level value of 170.

[0163] In this step, the deep learning demo with a gray level value of 170 is the defect detection model corresponding to the gray level range [140, 200] in the foregoing embodiment. When the gray level of the cell is less than 200, the processing unit 104 can use the deep learning demo with a gray level value of 170 to detect the image to be detected, obtain the detection result, that is, complete the detection and analysis of the cell quality based on the image processing algorithm corresponding to the gray level range [140, 200]; then enter step S12.

[0164] S11: Detect the image using the deep learning example with a gray level value of 210.

[0165] In this step, the deep learning demo with a gray level value of 210 is the defect detection model corresponding to the gray level range [180, 240] in the foregoing embodiment. When the gray level of the cell is not less than 200, the processing unit 104 can use the deep learning demo with a gray level value of 210 to detect the image to be detected, obtain the detection result, that is, complete the detection and analysis of the cell quality based on the image processing algorithm corresponding to the gray level range [140, 200]; then enter step S12.

[0166] S12: Output the detection result.

[0167] In this step, the detection result output by the processing unit 104 can be the foregoing prompt information output on the display screen, that is, display the processing result obtained through the vision algorithm on the display. Alternatively, it can also be the alarm information sent to the alarm 107 in the foregoing embodiment.

[0168] Based on the above processing, when the conveyor belt 101 conveys the battery cells to be detected to the detection area, the laser 102 automatically emits laser light towards the detection area, and the camera 103 automatically collects the image of the detection area under the irradiation of the laser light emitted by the laser 102, that is, collects the image to be detected of the battery cells to be detected under the irradiation of the laser light emitted by the laser 102. Subsequently, the processing unit 104 detects the image to be detected to obtain the detection result of whether there are defects in the battery cells to be detected. That is, the defect detection of the battery cells to be detected is automatically completed by the camera 103, the laser 102, and the processing unit 104, without the need for technicians to manually detect whether each battery cell is a defective battery cell, which can improve the efficiency of defective battery cell detection.

[0169] Moreover, in order to effectively perform defect detection, it is necessary that the gray values of the images of each type of battery cell be within the effective detection range under the irradiation of the laser. In the embodiment of the present application, for the laser light emitted by the laser 102, the difference in the light transmittance of any two different battery cells is less than a preset difference threshold, that is, the light transmittance of different battery cells is similar under the irradiation of the laser light emitted by the laser 102 in the present application, which can reduce the difference in the gray values of the images of different battery cells collected by the camera 103 under the irradiation of the laser light emitted by the laser 102. Furthermore, under the irradiation of the laser light emitted by the laser 102, the gray values of the images of each type of battery cell collected by the camera 103 are within the effective detection range, reducing the influence of the difference between different battery cells on the difference in the gray values of the collected images. That is, based on the defective battery cell detection system 10 provided in the present application, only one set of system is needed, and for different types of battery cells, the camera 103 can collect images to be detected with similar gray values under the irradiation of the laser light emitted by the laser 102. Subsequently, the processing unit 104 performs defective battery cell detection on the images to be detected with similar gray values.

[0170] Compared with the situation of a single process re-inspection station, due to the numerous process sections from the original silicon wafer to the pre-line of the battery cell, the texturing, diffusion, and coating processes may cause surface defects of the battery cell without destroying the internal structure of the battery cell, that is, it is necessary to use multiple cameras and multiple systems to detect the battery cells produced in different process sections, that is, multiple sets of hardware and multiple sets of software are required for supporting detection, and the cost of the solution is extremely high. Based on the defective battery cell detection system 10 provided by the present application, there is no need to set up multiple systems according to the type of battery cell, which can reduce the cost of defective battery cell detection in actual scenarios. That is, a linear array camera and a customized laser tube light source with widened line width and increased power are used to bring the average value of the light power transmitted by different battery cells closer, so as to achieve product compatibility. Using the same set of detection equipment for detection, it is compatible with the imaging effects of battery cells in the process sections from the original silicon wafer to the silicon nitride film coating, which reduces the complexity of the mechanism design, reduces the amount of hardware used, and reduces the difficulty and cost of system design. According to the output image effect, the needs of detecting hidden cracks in battery cells in the process section from the original silicon wafer to the silicon nitride film coating can be met, and fast and accurate defect detection can be achieved. Subsequently, the defect detection results of each battery cell to be inspected can be statistically analyzed to determine the source of the defective battery cell and the cause of the defect, so as to reduce production costs and increase battery cell production capacity.

[0171] Furthermore, cell types vary widely, and production volumes are high, resulting in numerous defects. Manual inspection can only identify cosmetic defects, such as perforations. Internal structural damage, such as hidden cracks, is difficult to detect through manual inspection, leading to poor defect consistency. This results in low manual inspection accuracy and high labor costs, which in turn reduces factory production capacity and prevents production yields from being met. The defective cell detection system 10 provided in this application can detect internal hidden cracks that are difficult to see with the naked eye. Furthermore, the laser is illuminated using a stroboscopic triggering method. Compared to a continuously on laser, this can extend the laser's lifespan and reduce safety hazards during laser operation. Furthermore, the algorithm software on the processing unit 104 determines the appropriate processing flow based on the cell's grayscale. The captured images are assigned to three processes corresponding to different grayscale values, which are then verified using three sets of deep learning demos. This system is compatible with all process stages, from raw silicon wafers to silicon nitride coating, for hidden crack re-inspection and rework inspection of cells.

[0172] Based on the same inventive concept, an embodiment of the present application also provides a defective cell detection method, which is applied to a defective cell detection system. The system includes a conveyor belt, a laser, a camera, and a processing unit. The laser is located on one side of a detection area within the conveyor belt, and the camera is located on the other side of the detection area. The defective cell detection system in this embodiment is similar to the defective cell detection system described in any of the above embodiments. For details, please refer to the relevant descriptions of the above embodiments.

[0173] See Figure 8 , Figure 8 which is a flowchart of a defective cell detection method provided by an embodiment of the present application. The defective cell detection method may include the following steps:

[0174] S801: When the cell to be detected on the conveyor belt is conveyed to the detection area, the laser emits laser light towards the detection area.

[0175] Among them, for the laser light emitted by the laser, the difference in light transmittance between any two different cells is less than a preset difference threshold.

[0176] S802: When the cell to be detected on the conveyor belt is conveyed to the detection area, the camera captures an image of the detection area under the laser light emitted by the laser to obtain the image to be detected; and sends the image to be detected to the processing unit.

[0177] S803: The processing unit obtains the image to be detected; detects the cell to be detected in the image to be detected to obtain a detection result on whether the cell to be detected has defects.

[0178] Based on the above processing, when the conveyor belt conveys the cell to be detected to the detection area, the laser automatically emits laser light towards the detection area, and the camera automatically captures an image of the detection area under the laser light emitted by the laser, that is, captures the image to be detected of the cell to be detected under the laser light emitted by the laser. Subsequently, the processing unit detects the image to be detected to obtain a detection result on whether the cell to be detected has defects. That is, the defect detection of the cell to be detected is automatically completed by the camera, the laser, and the processing unit, without the need for technicians to manually detect whether each cell is a defective cell, which can improve the efficiency of defective cell detection.

[0179] Moreover, in order to effectively perform defect detection, the gray values of the images of each type of battery cell need to be within the effective detection range under the irradiation of the laser. In the embodiments of the present application, for the laser emitted by the laser device, the difference in the light transmittance between any two different battery cells is less than a preset difference threshold. That is, in the present application, the light transmittance of different battery cells is similar under the irradiation of the laser emitted by the laser device, which can reduce the difference in the gray values of the images of different battery cells collected by the camera under the irradiation of the laser emitted by the laser device. Furthermore, under the irradiation of the laser emitted by the laser device, the gray values of the images of each type of battery cell collected by the camera are within the effective detection range, reducing the influence of the difference between different battery cells on the difference in the gray values of the collected images. That is, based on the defect battery cell detection system provided in the present application, only one set of system is needed, and for different types of battery cells, the camera can collect the to-be-detected images with similar gray values under the irradiation of the laser emitted by the laser device. Subsequently, the processing unit performs defect battery cell detection on the to-be-detected images with similar gray values. There is no need to separately set up multiple sets of systems according to the types of battery cells, which can reduce the cost of defect battery cell detection in the actual scenario.

[0180] In some embodiments, the camera is a line array camera, and the laser device is a line laser; when the to-be-detected battery cell on the conveyor belt is conveyed to the detection area, the laser device emits laser to the detection area, including: when the to-be-detected battery cell starts to enter the detection area, the laser device emits laser to the detection area at a specified frequency until the to-be-detected battery cell leaves the detection area; when the to-be-detected battery cell on the conveyor belt is conveyed to the detection area, the camera collects the image of the detection area under the irradiation of the laser emitted by the laser device to obtain the to-be-detected image, including: when the to-be-detected battery cell starts to enter the detection area, the camera collects the image of the detection area under the irradiation of the laser emitted by the laser device at the specified frequency until the to-be-detected battery cell leaves the detection area, and splices the collected images to obtain the to-be-detected image.

[0181] In some embodiments, the system further includes: an encoder; the encoder is connected to the transmission shaft of the conveyor belt; the specified frequency is the counting frequency of the encoder; when the battery cell to be detected starts to enter the detection area, the camera collects images of the detection area under the laser emitted by the laser at the specified frequency until the battery cell to be detected leaves the detection area, and stitches the collected images to obtain a to-be-detected image, including: according to the specified frequency, the encoder sends a collection instruction to the camera; when the battery cell to be detected starts to enter the detection area, the camera collects images of the detection area under the laser emitted by the laser according to the received collection instruction until a preset number of images are collected, and stitches the preset number of collected images to obtain a to-be-detected image; wherein, the preset number is determined based on the specified frequency, the conveying speed of the conveyor belt, and the lengths of various battery cells in the conveying direction of the conveyor belt.

[0182] In some embodiments, the system further includes: a photoelectric sensor; along the conveying direction, the sensing area of the photoelectric sensor is located in front of the detection area; when the battery cell to be detected starts to enter the detection area, the camera collects images of the detection area under the laser emitted by the laser according to the received collection instruction, including: when there is an object in the sensing area, the photoelectric sensor sends an arrival signal to the camera; after receiving the arrival signal, the camera collects images of the detection area under the laser emitted by the laser according to the received collection instruction.

[0183] In some embodiments, after receiving the arrival signal, the method further includes: when receiving the collection instruction, the camera sends a trigger instruction to the laser; when receiving the trigger instruction, the laser emits laser to the detection area.

[0184] In some embodiments, the battery cell to be detected in the to-be-detected image is detected to obtain a detection result on whether the battery cell to be detected has defects, including: inputting the to-be-detected image into a pre-trained defect detection model to obtain a detection result on whether the battery cell to be detected in the to-be-detected image has defects; wherein, the pre-trained defect detection model is trained using sample images containing sample battery cells; the sample images are collected under the laser emitted by the laser.

[0185] In some embodiments, inputting the image to be detected into a pre-trained defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the image to be detected, including: determining the gray value of the image area occupied by the battery cell to be detected in the image to be detected as the gray value to be processed; from the corresponding relationship between the preset gray value interval set and the defect detection model set, the processing unit determines the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs, and inputs the image to be detected into the determined defect detection model to obtain a detection result on whether there are defects in the battery cell to be detected in the image to be detected; wherein, each defect detection model in the corresponding relationship is obtained by training on images collected under the laser emitted by the laser for different types of battery cells; the gray value interval corresponding to a defect detection model represents the range of gray values of the images used to train the defect detection model.

[0186] In some embodiments, the gray value interval set includes three gray value intervals; the defect detection model set includes three defect detection models; the three gray value intervals and the three defect detection models are in one-to-one correspondence.

[0187] In some embodiments, for each type of battery cell, under the laser emitted by the laser, the light transmittance of this type of battery cell is not less than a preset light transmittance threshold; the exposure time of the camera is negatively correlated with the preset light transmittance threshold; in any image collected by the camera under the irradiation of the laser emitted by the laser, the gray value of the image area occupied by the battery cell belongs to a preset gray interval.

[0188] In some embodiments, the line width of the laser emitted by the laser is not less than 5 mm and not more than 12 mm; the laser power of the laser emitted by the laser is not less than 25 W.

[0189] In some embodiments, the system further includes an alarm; after detecting the battery cell to be detected in the image to be detected to obtain a detection result on whether there are defects in the battery cell to be detected, the method further includes: when the detection result indicates that there are defects in the battery cell to be detected, the processing unit sends an alarm message to the alarm; after receiving the alarm message, the alarm gives an alarm.

[0190] It should be noted that in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

[0191] Each embodiment in this specification is described in a related manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments.

[0192] The above description is only the preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are all included in the protection scope of the present application.

Claims

1. A defective cell detection system, characterized in that, The system includes: a conveyor belt, a laser, a camera, and a processing unit; the laser is located on one side of the detection area in the conveyor belt, and the camera is located on the other side of the detection area; wherein: The conveyor belt is used to convey the battery cells to be detected on the conveyor belt; The laser is used to emit laser to the detection area when the battery cell to be detected is conveyed to the detection area; wherein, for the laser emitted by the laser, the difference in light transmittance between any two different battery cells is less than a preset difference threshold; The camera is used to collect an image of the detection area under the irradiation of the laser emitted by the laser when the battery cell to be detected is conveyed to the detection area, to obtain a to-be-detected image; and send the to-be-detected image to the processing unit; The processing unit is used to obtain the to-be-detected image; detect the battery cell to be detected in the to-be-detected image, and obtain a detection result on whether the battery cell to be detected has a defect.

2. The system according to claim 1, wherein The camera is a line array camera, and the laser is a line laser; The laser is specifically used to emit laser to the detection area at a specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area; The camera is specifically used to collect an image of the detection area under the irradiation of the laser emitted by the laser at the specified frequency when the battery cell to be detected starts to enter the detection area until the battery cell to be detected leaves the detection area, and splice the collected images to obtain a to-be-detected image.

3. The system according to claim 2, characterized in that The system further includes: an encoder; the encoder is connected to the transmission shaft of the conveyor belt; the specified frequency is the counting frequency of the encoder; The encoder is used to send a collection instruction to the camera at the specified frequency; The camera is specifically used to collect an image of the detection area under the irradiation of the laser emitted by the laser according to the received collection instruction when the battery cell to be detected starts to enter the detection area until a preset number of images are collected, and splice the preset number of collected images to obtain a to-be-detected image; wherein, the preset number is determined based on the specified frequency, the conveying speed of the conveyor belt, and the lengths of various battery cells in the conveying direction of the conveyor belt.

4. The system according to claim 3, wherein The system further includes: a photoelectric sensor; along the conveying direction, the sensing area of the photoelectric sensor is located in front of the detection area; The photoelectric sensor is used to send an arrival signal to the camera when there is an object in the sensing area; The camera is specifically used to collect an image of the detection area under the irradiation of the laser emitted by the laser according to the received collection instruction after receiving the arrival signal until a preset number of images are collected, and splice the preset number of collected images to obtain a to-be-detected image.

5. The system according to claim 4, wherein The camera is further used to send a trigger instruction to the laser when receiving the collection instruction after receiving the arrival signal; The laser is specifically configured to emit laser light to the detection area when receiving the trigger instruction.

6. The system according to claim 1, wherein The processing unit is specifically configured to input the image to be detected into a pre-trained defect detection model to obtain a detection result indicating whether there is a defect in the battery cell to be detected in the image to be detected; wherein, the pre-trained defect detection model is trained using sample images containing sample battery cells; the sample images are captured under the irradiation of the laser light emitted by the laser.

7. The system according to claim 6, wherein The processing unit is specifically configured to determine the gray value of the image area occupied by the battery cell to be detected in the image to be detected as the gray value to be processed. From the corresponding relationship between the preset gray value interval set and the defect detection model set, determine the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs, and input the image to be detected into the determined defect detection model to obtain a detection result indicating whether there is a defect in the battery cell to be detected in the image to be detected. Among them, each defect detection model in the corresponding relationship is trained based on images captured under the irradiation of the laser light emitted by the laser for different types of battery cells; the gray value interval corresponding to a defect detection model represents the range of gray values of the images used to train this defect detection model.

8. The system according to claim 7, wherein The gray value interval set includes three gray value intervals; the defect detection model set includes three defect detection models; the three gray value intervals and the three defect detection models are in one-to-one correspondence.

9. The system according to claim 1, wherein For each type of battery cell, under the irradiation of the laser light emitted by the laser, the light transmittance of this type of battery cell is not less than a preset light transmittance threshold. The exposure time of the camera is negatively correlated with the preset light transmittance threshold; in any image captured by the camera under the irradiation of the laser light emitted by the laser, the gray value of the image area occupied by the battery cell belongs to a preset gray interval.

10. The system according to claim 9, wherein The line width of the laser light emitted by the laser is not less than 5 mm and not greater than 12 mm; the laser power of the laser is not less than 25 W.

11. The system according to any one of claims 1-10, characterized in that, The system further includes an alarm. The processing unit is further configured to send an alarm message to the alarm when the detection result indicates that there is a defect in the battery cell to be detected. The alarm is configured to give an alarm after receiving the alarm message.

12. A method for detecting defective solar cells, characterized in that, Applied to a defective battery cell detection system, the system includes: a conveyor belt, a laser, a camera, and a processing unit; the laser is located on one side of the detection area in the conveyor belt, and the camera is located on the other side of the detection area; the method includes: When the battery cell to be detected on the conveyor belt is conveyed to the detection area, the laser emits laser light to the detection area; wherein, for the laser light emitted by the laser, the difference in light transmittance between any two different battery cells is less than a preset difference threshold. When the battery cell to be detected on the conveyor belt is conveyed to the detection area, the camera captures an image of the detection area under the irradiation of the laser light emitted by the laser to obtain an image to be detected; and sends the image to be detected to the processing unit. The processing unit obtains the image to be detected, and detects the battery cell to be detected in the image to be detected to obtain a detection result on whether there is a defect in the battery cell to be detected.

13. The method according to claim 12, wherein The camera is a line array camera, and the laser is a line laser; When the battery cell to be detected on the conveyor belt is conveyed to the detection area, the laser emits laser light to the detection area, including: When the battery cell to be detected starts to enter the detection area, the laser emits laser light to the detection area at a specified frequency until the battery cell to be detected leaves the detection area; When the battery cell to be detected on the conveyor belt is conveyed to the detection area, the camera acquires an image of the detection area under the laser light emitted by the laser to obtain an image to be detected, including: When the battery cell to be detected starts to enter the detection area, the camera acquires an image of the detection area under the laser light emitted by the laser at the specified frequency until the battery cell to be detected leaves the detection area, and splices the acquired images to obtain an image to be detected; and / or The system further includes: an encoder; the encoder is connected to the transmission shaft of the conveyor belt; the specified frequency is the counting frequency of the encoder; When the battery cell to be detected starts to enter the detection area, the camera acquires an image of the detection area under the laser light emitted by the laser at the specified frequency until the battery cell to be detected leaves the detection area, and splices the acquired images to obtain an image to be detected, including: According to the specified frequency, the encoder sends an acquisition instruction to the camera; When the battery cell to be detected starts to enter the detection area, the camera acquires an image of the detection area under the laser light emitted by the laser according to the received acquisition instruction until a preset number of images are acquired, and splices the preset number of acquired images to obtain an image to be detected; wherein, the preset number is determined based on the specified frequency, the conveying speed of the conveyor belt, and the length of various battery cells in the conveying direction of the conveyor belt; and / or The system further includes: a photoelectric sensor; along the conveying direction, the sensing area of the photoelectric sensor is located in front of the detection area; When the battery cell to be detected starts to enter the detection area, the camera acquires an image of the detection area under the laser light emitted by the laser according to the received acquisition instruction, including: When there is an object in the sensing area, the photoelectric sensor sends an arrival signal to the camera; After receiving the arrival signal, the camera acquires an image of the detection area under the laser light emitted by the laser according to the received acquisition instruction; and / or After receiving the arrival signal, the method further includes: When receiving the acquisition instruction, the camera sends a trigger instruction to the laser; When receiving the trigger instruction, the laser emits laser light to the detection area; and / or Detect the battery cell to be detected in the image to be detected, and obtain the detection result of whether there are defects in the battery cell to be detected, including: Input the image to be detected into a pre-trained defect detection model to obtain the detection result of whether there are defects in the battery cell to be detected in the image to be detected; wherein, the pre-trained defect detection model is trained using sample images containing sample battery cells; the sample images are collected under the laser emitted by the laser; and / or, Input the image to be detected into a pre-trained defect detection model to obtain the detection result of whether there are defects in the battery cell to be detected in the image to be detected, including: Determine the gray value of the image area occupied by the battery cell to be detected in the image to be detected as the gray value to be processed; From the corresponding relationship between the preset gray value interval set and the defect detection model set, the processing unit determines the defect detection model corresponding to the gray value interval to which the gray value to be processed belongs, and inputs the image to be detected into the determined defect detection model to obtain the detection result of whether there are defects in the battery cell to be detected in the image to be detected; Wherein, each defect detection model in the corresponding relationship is trained based on images collected under the laser emitted by the laser for different types of battery cells; the gray value interval corresponding to a defect detection model represents the range of gray values of the images used to train the defect detection model; and / or, The gray value interval set includes three gray value intervals; the defect detection model set includes three defect detection models; the three gray value intervals and the three defect detection models correspond one by one; and / or, For each type of battery cell, under the laser emitted by the laser, the light transmittance of the battery cell is not less than a preset light transmittance threshold; The exposure time of the camera is negatively correlated with the preset light transmittance threshold; in any image collected by the camera under the irradiation of the laser emitted by the laser, the gray value of the image area occupied by the battery cell belongs to a preset gray interval; and / or, The line width of the laser emitted by the laser is not less than 5 mm and not more than 12 mm; the laser power emitted by the laser is not less than 25 W; and / or, The system further includes an alarm; After detecting the battery cell to be detected in the image to be detected and obtaining the detection result of whether there are defects in the battery cell to be detected, the method further includes: When the detection result indicates that there are defects in the battery cell to be detected, the processing unit sends an alarm message to the alarm; After receiving the alarm message, the alarm gives an alarm.

Images